1 Introduction
Interviewing is a primary way of collecting data in qualitative research to direct the participant in responding to a specific research question. Generally, three types of interviews are common: (1) structured; (2) semi-structured; and (3) narrative interview (Green and Thorogood 2018). The primary difference between these types is the amount of control the interviewer has over the encounter and the aim of the interview (Stuckey 2013). In traditional narrative interviewing, narratives are referred to as stories that are based on the unfolding of events or actions from the perspective of a participant’s life experience. However, in modern qualitative research, the assumption that participants' memories may simply be accessed through narratives that serves as a source of “data” is problematic. From this epistemological viewpoint, a narrative focus on both the told and the untold contribute to the constructive potential of the inquiry process (Gemignani 2014). It should be noted that resistance to dialog with the interviewer can be manifested not only by understatement and silence, but also by laughter and banter, which can be the subject of a separate analysis (Nairn et al. 2005). On the other hand, interviewees may overemphasize certain topics based on unwarranted assumptions about the researchers’ interest in them: that they are asked for an interview because they belong to a particular social or professional category (“interviewees with an agenda” according to Jacobsson and Åkerström 2013). Thus, they rather than talking to the interviewer are responding to or arguing against a “public view” or a “public discourse”. Another problematic aspect of interviews (especially when analyzing transcripts) is the propensity towards tensions that emerge—ambiguities, inconsistencies, contradictions (Watson 2006). Using these tensions in interviews enables narratives to be seen as a process of writing and re-writing in which the “interminable process of identification with” (Derrida 1998) is played out within the shifting discourses in which interview participants are immersed (Watson 2006). Taking into account the mentioned problematics, new approaches to analyzing subtexts are required.
A promising approach may be to use Configurational Comparative Methods (Rihoux and Ragin 2009), e.g. Qualitative Comparative Analysis (QCA). A detailed review of the challenges in converting qualitative into quantitative data in QCA was conducted by de Block and Vis (2019). In the study of Pagliarin et al. (2023), it was shown that QCA as a research approach is composed of (1) an “analytical move”, where cases, conditions and outcome(s) are conceptualized in terms of sets, and (2) a “membership move”, where set membership values are qualitatively assigned by the researcher (i.e. calibration). By using examples, the authors (Pagliarin et al. 2023) showed how researchers can gain access to in-depth casebased knowledge, or case intimacy, by adopting a “dialogical” interviewing style. However, Pagliarin et al. (2023) focused on the use of QCA “as a research approach” rather than “as a method”. From our point of view, the development of QCA in the analysis of narrative interviews “as a method” could be promising by using the formalized event model of the Russian-Estonian literary scholar and semiotician Juri Lotman.
Anzeige
In his publications (1977, 2003), Lotman conceptualized plot in terms of a notion of “event”. Even before Lotman, the concept of the event was commonly used to define narrativity in terms of the sequence inherent in the story being told. However, Lotman's view of an event differed from conventional ideas in that it did not refer to individual actions but to the global structure of the action depicted in a narration. According to Lotman (1977), a narrative plot needs at least three components: (1) some semantic field divided into two mutually complementary subsets; (2) the border between these subsets, which under normal circumstances is impenetrable, though in a given instance (a text with a plot always deals with a given instance) it proves to be penetrable for the hero-agent; (3) the hero-agent. As stated by Lotman, there is a narrative structure only if a text includes at least one “event”. For its part, an event occurs if and only if a character is displaced across the boundary between two spaces. Speaking of space, Lotman (1977) primarily had in mind its semantic aspects as evidenced by the following phrase: „…an event always involves the violation of some prohibition and is always a fact which takes place, though it need not have taken place. For a man who thinks in terms of the categories of the criminal code, an event is an act of transgression…”.
The first formal logic formulation of Lotman’s event model was that by Renner (1983, 2004). The special feature of Renner's formulation was that it was carried out within the framework of set theory with a transition to propositional logic. Renner understood events to be disruptions of order, and disruptions of order in turn as logical contradictions. In addition to the formal logic representation of disruptions of order, Lotman and Renner differ in that Renner presupposes that the “eventful situation” arises and continues, i.e. that the eventful situation continues until the contradiction is once more resolved. However, Renner’s definition of the event as the arising of a logical contradiction and of the eventful situation as the continuation of that logical contradiction produces a problem: formal logic excludes the possibility of a contradiction because of the principle of non-contradiction. This follows from the fact that, in formal logic, a contradiction must always be resolved. The occurrence of the event, therefore, demands its immediate termination—either through a change of the ordering statements which determine how a character should be or behave or through a change of the description of the situation which determines how a character should be or behave. That, however, contradicts Renner’s understanding of the durative aspect of the event. The consequences of the principle of non-contradiction do not occur to Renner. He sees the problem in a different circumstance: his ordering statements lack the characteristics of postulates. It can be deduced from his writings that he wishes his ordering statements to be understood as normative statements. However, non-modal sentences, and these are precisely what Renner uses, do not describe, even in the case of subjunctions/implications, what should be, but merely what is. Renner’s ordering statements, therefore, are equivalent in formal logic to statements of description, with the sole difference being that the ordering statements, because of their if–then form, are conditional actual-state statements: if x applies, then so does y. This is in no sense a normative statement. Renner, then, sketches out a model using ordering statements with the operators of necessity of modal logic, without being able to solve the problem he had correctly recognized.
A possible solution to these problems was proposed by Klimczak (2016) based on of normative reasoning/deontic modal logic. Instead of an operator of necessity, an operator of obligation was used. Using this operator, it was possible to produce contradiction-free statements about how something should be, which could be placed in relation to selective statements about how things were. This deontic formulation of Lotman’s event with an ontic extension by the model of the possible worlds resulted in the method called Formal Subtext Analysis (FSA) which was initially conceived for the study of narrative plots of fictional texts (Klimczak 2014, 2016).
This study aims to demonstrate this method for the deduction of subtexts and to discuss its application as an extension of QCA in narrative interviewing.
Anzeige
The main content of the paper is divided into three sections. In the theoretical section, we first describe a deontic remodeling of Lotman's event; in the second, we give its ontic extension using a modal logic model of possible worlds; and in the third, we demonstrate the potential for a subtext deduction. In the practical section, we present an example of narrative subtext analysis based on an interview with a doctor from a mental health hospital in Germany. Finally, the possibilities for the development of the presented model as an extension of QCA are discussed.
2 Formal logic remodeling of Lotman’s event model
2.1 Deontic remodelling of the event
In his book “Elements of Narratology” which was inspired by works with former members of the Tartu-Moscow Semiotic School, the German narratologist Wolf Schmid (2014) introduces five criteria of narrative eventfulness. These are: (1) relevance (extent of how a change of state in the respective narrative world is perceived as essential), (2) unpredictability (extent of deviation from the narrative “doxa”, the general expectation), (3) consecutiveness (extent to which a change in the context of the narrated world has consequences for the thoughts and actions of the subject concerned), (4) irreversibility (extent of the unlikelihood that the new state resulting from the change will be reversed) and (5) non-iteration (extent of the uniqueness of a narrative change). Regarding unpredictability, Schmid found this criterion to be broader than only “the violation of some prohibition” described by Lotman. Considering that in his later works Lotman applied the analytical methods developed in the context of his philological studies to broader cultural material (the world as a text), then Schmid's criterion for the unpredictability of events is undoubtedly given in Lotman's “The Unpredictable Workings of Culture” and “Culture and Explosion”. However, in our concept, we use terms of deontic logic close to Lotman's early understanding of events. Generally, we assume all five criteria to be consistent with Lotman's thinking and therefore include them in our formal logic considerations as useful terms.
The subject of deontic logic is the logic of obligation, prohibition and permission. A logical calculus is created by adding operators “O(p)” (it ought to be the case that p) and “P(p)” (it is permissible that p) to a basic system such as the propositional calculus, together with rules of inference for them. The forbidden operator “F(p)” is the logical equivalent to “O(¬p)” and is usually replaced by it. “While one can say using O(p) that a certain state of affairs, namely p, ought to be, ¬p states that p does not apply. Since it does not apply that the being of p follows from the oughtness of p, a formal-logical contradiction between an “is” statement and an “ought” statement cannot be deduced. Since the operator “O(p)“ expresses a kind of moral obligation that corresponds to the general expectation, its violation can be assumed as highly unpredictable. An event can then be built according to Lotman as an equivalent to a “is” ¬p and an “ought” p, O(p), which can be formalized: E↔¬p∧O(p), i.e. “An event occurs if and only if not-p applies and it ought to be that p applies”. Using this event formula, it is now possible to deduce the existence or non-existence of events without the deduction of contradictions: [¬p∧O(p)]∧(E↔¬p∧O(p)]→E and [p∧O(p)]∧(E↔¬p∧O(p)]→¬E. Conversely, it is also possible to deduce conclusions about the ought” statement on which an event is based: (E∧¬p)∧[E↔¬p∧O(p)]→O(p) as well as the conclusion that a certain “is” statement is not valid, in the case of presence of a non-event condition: (¬E∧¬p)∧[E↔¬p∧O(p)]→¬O(p).
The relevance of an event is its substantial characteristic and therefore cannot be represented directly using formal logic means. Nonetheless, in most cases, something is only desired or ought to be under a relevant condition. However, in deontic modal logic, there is disagreement about how conditional “ought” statements are to be formalized. For deontically determined “ought” statements, the form is usually used O(q→p), i.e. “It ought to be that p applies if q applies”, or for material conditional “ought”, the form q→ O(p), “If q applies then it ought that p applies”. In the context of Lotman's event model, however, the use of deontic ought-statements would inevitably lead to paradoxes and logical contradictions (Klimczak 2016). The use of material conditional “ought” solves this problem, but it brings with it the new problem that material conditional “ought” (in contrast to deontically conditioned ones) cannot be operationalized using the event formula (Klimczak 2016). A corresponding extension for conditional ought sentences is therefore required: E↔ (q∧¬p)∧[q→O(p)], i.e. “An event occurs if and only if q and not-p applies and if q applies, it ought that p applies”. In contrast to secondary “ought” statements, which are described in Sect. 2.3 we call expressions like q→O(p) primary “ought” statements.
While non–iteration and irreversibility criteria were introduced based on the analysis of literary texts, the narrative texts of qualitative interviews in some social and political areas may have substantial deviations from them. For example, a conspiracy theory is defined as a representation that explains a series of events by postulating a conspiracy as its cause; that is, the events are seen as “the result of a group of people acting in secret to a nefarious end” (Birchall 2006; Leone et al. 2020). Accordingly, the recurrence of events will lead to an increase in eventfulness. From the semiotic perspective, some political events can also be successfully reversed, e.g. “the undoing of an event” (Žižek 2014). However, our event remodeling allows the iteration and reversibility of events, without this leading to logical contradictions.
The consecutiveness criterion of eventfulness did not flow into the presented remodelling scheme as a syntactic element. Taking into account the object of our study—texts of qualitative interviews, we considered it to be sufficient to represent the subjective consequences of an event using formal logical deduction of a narrative subtext (see Sect. 2.3).
2.2 Ontic extension of Lotman’s event model
Even independently of the representation of consecutiveness, narrative texts are characterized by at least one feature change, which, however,—even within the framework of a deontic modeling—brings a formal-logical contradiction. The solution to this problem is that with each feature change a new point in time of the represented world is assumed and the individual points in time are understood as possible worlds in the sense of ontic modal logic (Klimczak 2014, 2016). According to Kripke (System S5) (1959), “possible worlds” are technical terms from the field of formal logic that should not evoke any associations to parallel universes from the field of science-fiction literature.
In the framework of possible worlds, “ought” statements must be modeled as true at every point in time, which in turn means that they are to be assumed as necessarily true. By means of the necessity operator □, unconditional and conditional “ought” statements are to be represented as □[O(p)], i.e.”It is necessary that p is ought “ and □[q → O(p)], i.e. “It is necessary that p is ought if q applies”. This required a corresponding modification of the event formulas: E ↔ ¬p∧□[O(p)] and E ↔ (q∧¬p)∧□[q → ¬O(p)]. Since the represented world can be modeled as a finite sequence of individual points in time and the individual points in time can be modeled as possible worlds, formally everything that is not given at any point in time ¬(p∧q) is not necessary: □¬(p∧q). According to De Morgan’s second law, the expression □¬(p∧q) is the logical equivalent to □(¬p∨¬q), i.e. “t is necessary that either not-p applies or not-q applies, or both not-p and not-q apply”. Necessary disjunctions are themselves logically equivalent to necessary subjunctions: □(¬p∨¬q) ≡ □(p → ¬q), which can be recorded as condition relations (Klimczak 2014, 2016). If we then assume that ¬p and ¬q are predicates of an individual named X, then in the framework of the New Theory of Reference (Kripke 1980), □(¬Px∨¬Qx) resp. □(Px → ¬Qx) is an object denoted by corresponding rigid designators in all possible worlds.
In this context, the question arises: How close is the theory of possible worlds to Lotman's thought? In the article “Culture as a Subject and Its Own Object” (2019), Lotman turned to Leibniz's monadology, believing that the semiosphere consists of many “semiotic monads” as meaning-making entities.The fact that Lotman was very close to the metaphysics of possible worlds of Kripke was pointed out by Zenkin (2008):
According to Zenkin (2008), Lotman, in his paper “The Origin of Plot in the Light of Typology” (1979), focused on the phenomena of paradigmatic equivalence, which in the linguistic structure of the narrative are projected onto its syntagmatics (with the duplicity of characters, etc.), but on a deeper level take place in the sphere of reference.The rigid designator points to the inherent self-identity of an object that it retains when varying any of its other attributes, in any ‘stories’, where s/he could participate. In such an interpretation, the proper name is not a tool of semantics, but of ‘direct reference’, not mediated by any descriptions; in the words of another analytical philosopher, J. Searle, this is something like ‘pegs on which to hang descriptions’. Interestingly, Kripke also mentions a situation in which, in a possible world, the proper name of a character may change. Using as an example the figure of a President of the USA, Kripke remarks (1980): ‘…proper names are rigid designators, for although the man (Nixon) might not have been the President; it is not the case that he might not have been Nixon (though he might not have been called’Nixon’)’. In other words, a person, and in general any singular object that can be called a proper name, some ‘purely logical’ name is known, which in principle does not coincide with its linguistic name and contains the potential of its referential identity.
2.3 Deduction of the subtext(s)
Based on the last step described in the previous section, e.g. necessary subjunctions □(¬p∨¬q) ≡ □(p → ¬q), and using primary event-inducing “ought” statements, secondary “ought” statements can now be deduced using deontic modus ponens or tollens (Klimczak 2014, 2016): □[O(p)]∧□[□(p → q)] → □[O(q)] or □[O(¬q)]∧□[□(p → q)] → □[O(¬p)]. The starting point for the deduction of the secondary “ought” statements, in turn, represents a temporally differentiated feature-individual assignment, meaning a predicate-logical recording of individuals and predicates in the form of “is” statements. This assignment provides the basis for FSA according to Lotman's event model and is therefore a prerequisite of it. It is important to consider that an increase in the total number of predicates and/or a decrease in the number of time points represented as possible worlds lead to an increase in potential subtexts. A diversity of subtexts is also to be expected in the analysis of a set of interviews.
Substantially, a subtext as “a subliminally communicated opinion or message” (Brockhaus-Enzyklopädie 1993) can be expressed in the qualitative interview along a continuum ranging from an underlying communication to the interviewer (the I-s/he system) to an autocommunication, where the subject is transmitting a message to itself (Lotman 1990). The latter case is especially important for qualitative research since the initiative of the qualitative interview comes from the interviewer, who is just interested in learning someone else's opinion, regardless of his own person.
As stated by Lotman (1988), a subtext can be read from at least two viewpoints, one according to Propp and the other according to Bakhtin:
Whereas Propp’s method was oriented toward putting together the identical code text underlying different texts, representing them as a set of variations on one text, Bakhtin’s method, […], was the opposite: not only different but also—which is especially important—specifically untranslatable subtexts are discriminated in a single text. The internal conflict in a text is revealed. In Propp’s description, a text gravitates toward panchronic equilibrium: because narrative texts are examined, the fact that there is essentially no movement is especially conspicuous—there is only an oscillation around some homeostatic norm (equilibrium-disruption of equilibrium-restoration of equilibrium). In Bakhtin’s analysis, the inevitability of movement, change, and destruction is latent even in the static state of a text. Hence, a text is thematic even when it seems to be very far removed from the problems of theme and plot.
3 Empirical demonstration of FSA
3.1 Narrative interview
To clarify the model outlined above, a transcribed text of a narrative interview was used. This interview was conducted as part of the study “Subjective meaning qualities of experiencing violence in the context of mental health care and associated risk factors” that is being conducted in the Department of Psychiatry and Psychotherapy, Brandenburg Medical School, Immanuel Clinic Rüdersdorf, Germany. The study refers to the attachment theory of Bowlby (1979). According to Bowlby, people are born with a psychobiological system (the attachment behavioral system) that motivates them to seek proximity to supportive others (attachment figures) in times of need for the sake of gaining a sense of safety and security.
Using a preliminary recruitment survey, several employees of the Department of Psychiatry were selected who reported a change in the relationship with a psychiatric patient after a case of perceived aggression and the need for psychological elaboration of the situation. The qualitative interview presented here was conducted with a doctor in training to become a specialist in psychiatry. The interviewers were two doctors (RE and YI) who formally belonged to the same department but to different clinics which are located in different towns, as a result of which direct contacts between the interviewee and the interviewers were only episodic. Neither the interviewee nor the interviewers occupied a subordinate position to the other. Both interviewers had experience in conducting qualitative research, including interviews. Due to the spatial distance, the interview was conducted by phone in two parts, each lasting about 30 min. The first part was conducted by RE. The second part of the interview was conducted by YI and took place a few months after the first part to validate its results. After recording, transcription and pseudonymizing, the text of the interview and the first draft of the present paper were submitted to the interviewee for a critical overview and approved by him for publication.
At the time of the interview, the interviewed doctor (= David, pseudonymized name), who was born in Germany, was in his mid-thirties. Prior to the interview, in addition to general information about the study, he was briefed on the underlying attachment theory. However, he was not informed about the details of narrative analysis and the relevant predicates studied. Only those excerpts that were important for the narration or the analysis are presented in the following transcript.
After a short welcome phase, the interviewer (RE) started with a question as to whether David was able to talk about violence in psychiatry with reference to a self-experienced situation that was particularly memorable.
David began his story by saying that since he works in the external clinic of the psychiatric department, he himself was rarely a victim of violence, although he is well aware of cases that have taken place. However, he had to deal with violence, albeit rarely, while on on-call shifts in the main department. Further, he described himself in the following words:
Although at the very beginning David only wanted to talk about a single case, the narration was then spontaneously extended to two other cases. The first situation with an acutely psychotic patient happened a few weeks ago before the interview.… from my basic attitude I am actually always relatively erm I’d say trusting, so, as far as patients are concerned and actually always go very close in contact, so, erm, I am not shy at all (1). And so far, it has always been the case that my feeling has told me or have … trusted my feeling, erm, that nothing … that nothing would happen to me (2).
In the short inquiry phase after the free narration, David was able to access some additional details about what happened.… it was about a patient whom I am … that is a patient who has been in treatment for a long time, on the ward (3). I don't know him from the outpatient setting, but I only know him from night shifts and have often talked to him (4). He is in treatment because of a rather long-lasting psychosis and erm is also always so fluctuating with the symptoms, eh, but is overall such a very … yes somehow … so intuitively he was very sympathetic to me at first, although I knew that he can be very moody …(5). And I didn't talk to him about psychiatric things, but about music (6). The Beatles (7). Because he somehow … I think he kind of listened to music and then somehow we came to the conclusion that our favorite, that we both think the Beatles, at least the later Beatles, are really cool and which albums we like best (8). That's what we talked about (9). And that, whenever we saw each other again, sometimes on the ward, then we greeted each other briefly and … and named a song title by Beatles, somehow (10). And I always found that quite nice, even though I knew that the patient is sometimes also disliked by other colleagues because he is actually sometimes a bit abusive… (11). Yes and erm … but that's how I always had the feeling from him, erm, that we somehow have this kind of connection with each other, erm (12). And then that was just … yes, now a few weeks ago, that he …so there I had also noticed that he had worsened anyway with his symptoms, that he has become more aggressive, and then it was just, like, in the middle of the night that he was very erm … very upset and also … could not be argued with at all so also erm so the … was loud on the acute ward, the others could not sleep, you could somehow not find a compromise with him and he was then there in the smoking room and I only came in there to him … (13). And after I realized that he does not want to talk to me at all I got louder and tried to reprimand him and then he first approached me and he stood there so close in front of me (14). Well, I didn't go back at first but was afraid that something would happen, but at some point, I also went out of the room (15). I noticed it wasn’t leading anywhere and then he was louder and erm … so he was then quieter and then he was very loud again and then I went back to the smoking room erm (16). Well and then he just ran full towards the door and just kicked the door erm, erm with full force (17). Well, so that almost the [tempered] glass in the door broke and I just stood right in front of the glass (18). Erm, and if the glass in the door hadn’t been between us, then he would have kicked me in the stomach (19). Ermmmm … yes and that was actually such a violent experience of which I thought at first, yes, did not hit me, so in that sense it was not physical violence now, but I have to say, that actually stuck with me for the next few days (20). I also really winced and I just noticed that he would have given me a full blow (21). ….Erm and …erm well, so or that somehow still moves me now when I tell that…(22). […].… I came back (23). I went to the nurses (24). To the nurse station, into the service room and then I reported it so briefly and said, guys, I think you can't get any further with him, you can't try any further (25). And every further attempt can lead to escalation and then we more or less decided together that we will now leave him there in this room in his anger and don´t further … don´t further approach him and if it does not escalate further from his side, that we would not somehow further … further, or else call the police, or something (26).
Given this, the second attempt at communication was justified by the hope of bringing a personal component into the situation.So erm when I visited him for the first time erm in the smoking room, erm tried to calm him a bit …, so to calm him down …, or to help him regulate himself (27). There was erm still a nurse with him, a female one (28).
When asked what made this situation so special, David replied:Well, I somehow thought, I … felt somehow and went there on my own, just to try again in a confidential way (29). I thought, if I come with several people in the squad, it is already an intimidating attempt for him from the outset or at least he experiences it as such (30). Erm, that might arouse resistance in him (31).
Due to the persistent deterioration of the patient's mental state, no discussion with him about the situation could take place. David explained the patient's aggression as a wave of general anger that did not arise directly from their interaction, but that he was only a placeholder for this anger. However, since David came into contact with this patient again during subsequent on-call shifts, he could record changes in his own behaviour.Well, I think what, so … what was special about it was for me that… (32). Well, I think it was that closeness (33)! So, this closeness both in the sense of, he was just physically very close to me and just very violent, so… so that strong kick, and also the bang there, still… erm to feel it right in front of me, and to know that only the glass in the door had just saved me (34). And…and on the other hand, again, I think, that closeness I think because I already had identified with this …this patient … because I have identified with him to a certain extent, and have also felt him this close or something (35). This certainly has something to do with the social milieu (36). So, this is a relatively educated patient (37). So, he also somehow started to study and … and also like, considering his taste and so, habitus, he was … he was not so far away from me (38).
As an illustration of this, David spontaneously referred to the second situation with another problematic patient.….and since then I have just become a bit more cautious with this patient erm a bit more objective I would say, so … so this fooling around with each other, the erm …uhm I don't do that anymore (39). But I also have to say, this is still … so this … this greater restraint is limited to him, so I have not changed anything about my other …my other behavior (40).
Returning in the interview to the reflection on the mental consequences of the first situation with the acutely psychotic patient, David also spontaneously came up with a third final case. However, in his opinion, this situation was even more lasting than the first case described.… So, we have a patient on a ward, who for example … who is erm … has an intellectual disability and he is also sometimes very upset and aggressive and I remember once there he was also like that and then we had already got the restraint bed and erm … and I was also told, just do not go into the room, he beats you right away and pulls you over (41). And then I went to him and just naively asked what was going on and if he needed anything, and then we talked briefly and then I asked him if he might want a cold shower because he is so upset (42). And then he got a cold shower from me, so (laughs) went into the shower room together, and after that everything was fine and he did NOT need the restraint (43). And outside everyone was standing and thinking, oh, oh God, he's sitting here right now, pinching right now (44).
David described at the very end of the interview what affected him particularly.Well, but I have to say, so if I tell you this now, I once had … I once had a confrontation with a patient […] in group therapy (45). Well, above all I have to say, those very educated patients who sometimes have … who have the ability, because of their education, erm, to hurt you in a completely different way than by um using any swear words, or something (46). So erm, who sometimes wrote me such nasty, nasty emails (47). So, I think… I also felt that was a form of violence (48). Erm, he really tried to analyze and devalue certain personality traits of mine (49). That is also something… I got to say, something that still touches me today (50). That´s also a patient who is in our outpatient clinic… in fact … I also said, I don't want him to come to group therapy anymore (51). Erm, he is still being treated here by a colleague (52). I sometimes see him in the waiting room and then I'm a little bit …, because I notice that it affected me (53). So, erm, I have to say, that's also … well, it's also a form of violence that has left some lasting traces (54). And that also has something to do with closeness, I think, because the erm … because…, also had the similar… like a similar habitus simply, as far as now erm so social belonging, milieu affiliation is concerned (55). Erm, and because he frankly, did this very well …. in the sense of what he wanted to achieve, he achieved: I was broken (56). Yes (57). (long pause).
Since the second interview was based on the results of the analysis of the first, excerpts from the second interview are given in the following paper section.…well, he said something like this: “You are for me the personification … You are for me in your way and all your actions the personification of well-meant and badly done”, or something like that (58). And then he listed it like that with examples, and so on (59). And with that, he pulled everything… all my aspirations, this special group for people with mental disorders; he pulled everything, into the dirt (60).
3.2 Formal subtext analysis of the interview
3.2.1 Determination of the relevant characteristics/predicates
Since the characteristics of the described patients cannot be precisely tracked during narration, only features of David were included directly in the analysis. For didactic reasons, only three a priori defined features of David understood as processes are considered:
Emotional Approximation, in our context would be identification as a “term for a socio-psychological process in which such a strong connection between two people is experienced that under certain circumstances a mutual replacement of individualities becomes possible” (Hoffmeister 1955). “Individualities are manifested in personality features (personality traits), attitudes, interests and value orientations (values), religious, philosophical and political beliefs (belief system), in self-concepts (self-image), in social behavior (social interaction) and communication style (communication)” (Dorsch et al. 2014).
Physical Approximation, in our context would be: “…trusting behavior, that (a) increases one’s vulnerability, (b) to another whose behavior is not under one’s control, (c) in a situation in which the penalty suffered from abuses of that vulnerability is greater than the gains from compliance”(Deutsch 1962).
Physical or emotional closeness to the perpetrator was previously described as important component of the exposure to perceived aggression, essentially determining its experience (Deković et al. 2008; Vieselmeyer et al. 2017). The third feature is important due to the medical context of the study.Medical Intervention, in our context: any active form of treatment that can be distinguished from mere waiting. In the narrower sense, intervention means an acute, urgent intervention against a disease process. The intervention can be carried out, for example, as a surgical intervention (surgery), as a psychotherapeutic or as a drug intervention (conservative therapy).
In the language of predicate logic, David's emotional (E) and physical (P) approximation to the patient, as well as the medical intervention (I) performed by him, were designated as Ed, Pd, Id, respectively (“d” means David). While the negation of the first two processes ¬Ed, ¬Pd could represent both intentional and unintentional cessation, David's negation of medical intervention ¬Id could be intentional or involuntary. A subjective indicator of an event was David's feeling of shock defined as a surge of strong negative emotions (Sd). Since the focus was on David's processes rather than that of the patients, and in addition, all described patients had similar problem characteristics, the analysis did not distinguish between individual patients.
3.2.2 Secvenation of the transcribed text
The secvenation was carried out by identifying the individual non- and event-relevant points in time that corresponded to the possible worlds (pws). The sequence boundaries were determined by changing at least one relevant characteristic compared to the previous period:
-
pw = 0 (sentences 1–12): Ed, Pd, ¬Sd
-
pw = 1 (sentences 13–15, 27–28): ¬Ed, ¬Pd, Id, ¬Sd
-
pw = 2 (sentences 16, 29–31): Ed, ¬Pd, ¬Id, ¬Sd
-
pw = 3 (sentences 17–22): Ed, Pd, ¬Id, Sd
-
pw = 4 (sentences 23–26): ¬Ed, ¬Pd, ¬Id, ¬Sd
-
pw = 5 (sentences 39–40): ¬Ed, ¬Pd, Id, ¬Sd
-
pw = 6 (sentences 41–44): ¬Ed, Pd, Id, ¬Sd
-
pw = 7 (sentences 45, 58–60): Ed, Pd, ¬Id, Sd
-
pw = 8 (sentences 51–53): ¬Ed, ¬Pd, ¬Id, ¬Sd
Since no medical intervention was intentional or involuntary in pw = 0, which means that the predicate (Id) could not be tracked, pw = 0 was not included in the further analysis. In pw = 1, no physical approximation was possible, because a nurse was present. David wanted to rebuke the patient, which contradicts an emotional approximation. Nevertheless, the emotional approximation was developed in pw = 2 when David was in the service room. In short, in pw = 3 came the trusting behavior, which in perspective also required medical intervention. However, the intervention could not be realized and the combination of predicates in pw = 3 was event-inducing (Sd). While in pw = 3, the negation of medical intervention (¬Id) was involuntary, in pw = 4, it was a targeted action. Subsequent contacts with the patient in pw = 5 were medically motivated (Id) and free from emotional and physical approximations (¬Ed, ¬Pd). Although the contact with the intellectually disabled patient in pw = 6 did not exclude empathy, this also did not require an exchange of values (¬Ed). Emotional and physical approximation in pw = 7 were marked as positive because they are essential prerequisites of group therapy. However, during the confrontation no medical intervention was possible and the combination of predicates here could be defined as event-inducing (Sd). The confrontation ended in the cessation of all contact with the patient (¬Ed, ¬Pd, ¬Id), which was presented in pw = 8. Summing up, only pw = 3 and pw = 7 can be defined as event-inducing.
3.2.3 Deduction of primary “ought” statements for event-inducing and no event-inducing possible worlds
Since on the superficial level each event situation takes place within the framework of objective interactions between David and the patient, only predicates that could be traced by all stakeholders in objective reality (Pd, ¬Pd, Id, ¬Id) were included in this deduction. On the basis of potentially event-inducing predicates, all possible unconditional and material conditional “ought” statements for each event-inducing possible world were listed. Since these statements must be modeled as true in every possible word, they were to be assumed as necessarily (□) true. The deduction of the eligible unconditional and material conditional “ought” statements for each no event-inducing possible world was carried out according to the same principle. Due to the no event-inducing of several worlds, the statements were to be assumed as not necessarily (¬□) true. All “ought” statements according to the possible worlds are presented in Table 1.
Table 1
Distribution of unconditional and material conditional “ought” statements according to possible worlds
Non-event-inducing possible worlds | Event-inducing possible worlds | ||
|---|---|---|---|
pwa = 1, pw = 5 | pw = 2, pw = 4, pw = 8 | pw = 6 | pw = 3, pw = 7 |
¬Pdb ∧ Idc ∧ ¬Ed | ¬Pd ∧ ¬Id ∧ ¬E | Pd ∧ Id ∧ ¬E | Pd ∧ ¬Id ∧ E |
¬□[O(Pd)] | ¬□[O(Pd)] | ¬□[O(¬Pd)] | □[O(¬Pd)] |
¬□[O(¬Id)] | ¬□[O(Id)] | ¬□[O(¬Id)] | □[O(Id)] |
¬□[Id → O(Pd)] | ¬□[¬Id → O(Pd)] | ¬□[Id → O(¬Pd)] | □[¬Id → O(¬Pd)] |
¬□[¬Pd → O(¬Id)] | ¬□[¬Pd → O(Id)] | ¬□[Pd → O(¬Id)] | □[Pd → O(Id)] |
While only four “ought” statements could be deduced, twelve negations could be derived from “ought” statements. It was crucial that two of the four “ought” statements had to necessarily be excluded by the derived negations, i.e. were invalid. These were the two unconditional “ought” statements □[O(¬Pd)] and □[O(Id)], which were in contradiction with statements ¬□[O(¬Pd)] and ¬□[O(Id)], respectively. Only the two remaining conditional “ought” statements □[¬Id → O(¬Pd)] and □[Pd → O(Id)] were therefore to be assumed to be valid. Already with a preliminary inspection of these two remaining “ought” statements, it was clear that expression □[Pd → O(Id)] with O(Id) in the consequent was more likely to come into question as event-inducing, because doctors generally regard medical interventions as mandatory. However, this statement only covered those situations where medical intervention is appropriate (cf. in contrast pw = 0).
The full reading of the statement Pd → O(Id) would be: “David is supposed to perform a medical intervention if he behaves trustingly with the patient”. The causal connection between trusting behavior and medical intervention could be made clear in the second interview. To the question “What do you expect from the patient if you behave trustingly towards him?” David replied with regard to problematic situations:
When asked, “What would it mean if the patient didn't give these signals?” David replied:The insight, the openly shared feelings that are communicated […] by the patient saying: “I'm now erm, erm worse or better and I would like to discuss this with you” and […] for example that he says in the evening, “ok, I'm now ready to lie down in bed first, to calm down a bit (61). And if I need help again, then I’ll contact you or other staff again” (62). So that the patient communicates his willingness (63). […] The willingness to communicate his feelings, verbally, as well as a willingness to get help from us for these feelings, through a doctor's consultation or through medication or through perhaps also a contact with relatives (64).
However, in the first and third situations experienced by David, both patients not only did not give the expected signals but also devalued his offer of medical intervention brutally.Then it would unsettle me because then I would not know … or no feel … that's just intuition… then I would have a bad sense of how the night or erm the next hours would pass for example […] and so a bit more of an unpredictability maybe, what could still happen… (65).
3.2.4 Deduction of secondary “ought” statements (subtexts)
After reconstructing the event on the superficial level, a more in-depth analysis of the relations underlying the narration became possible. For this purpose, all combinations of the potentially event-inducing predicates whose relations have not yet been examined were listed. The relations of predicates or their negations, for those “ought” statements that were already identified in the previous step, should not be included in the analysis due to redundancy. In this respect, only the relations between Pd/¬Pd and Ed/¬Ed as well as between Id/¬Id and Ed/¬Eds were the subject of the examination (Table 2).
Table 2
Distribution of predicate relationships underlying the narration according to possible worlds
Combination of predicates | ||||
|---|---|---|---|---|
Pda | ¬Pd | Idb | ¬Id | |
Edc | Ed∧Pd | Ed∧¬Pd | E∧Id | Ed∧¬Id |
pwd = 3, pw = 7 | pw = 2 | pw = 3, pw = 7 | ||
¬Ed | ¬Ed∧Pd | ¬Ed∧¬Pd | ¬Ed∧Id | ¬Ed∧¬Id |
pw = 6 | pw = 1, pw = 4, pw = 5 pw = 8 | pw = 1, pw = 5, pw = 6 | pw = 4, pw = 8 | |
Regarding relations Ed/¬Ed and Pd/¬Pd, all four possible combinations in the narration were given. The situation was different in the case of Ed/¬Ed and Id/¬Id. The relation Ed∧Id could not be observed in any possible world, so it was necessarily not present: □[¬(Ed∧Id)]. However, according to De Morgan's second law, the relation □[¬(Ed∧Id)] is equivalent to □[(Ed∨Id)], which in turn could be converted to the material conditional □[(Ed → ¬Id)] resp. □[(Id → ¬Ed)]. Using deontic modus ponens, on the basis of □[(Id → ¬Ed)] and the primary “ought” statement □[Pd → O(Id)], a secondary “ought” statement could then be deduced: {□[□(Id → ¬Ed)]∧□[Pd → O(Id)]} → □[Pd → O(¬Ed)]. That is: “David should not approach the potentially aggressive patients emotionally if he behaves trustingly toward them.”
Although the identified “ought” statement could seem reasonable and justified from a non-academic point of view, it would sound very counterintuitive to a psychiatrist. Beginning with Jaspers' interpretive psychology (1923), it was an immutable truth in psychiatry that empathic understanding opens the door to the communication with the patient. Taking the same view, David however, did not rule out another possibility:
However, unlike emotional approximation, trusting behavior unfolds in objective reality. Thus, if emotional approximation should cause a greater degree of trustful behavior, then it can also increase vulnerability to external aggression. On the other hand, when asked whether trustful behavior without emotional approximation could be a profitable strategy, David admitted this, although he evaluated such a strategy as ethically ambiguous and associated it with possible punishment. However, the solution to the dilemma was also there:It´s a circuit that goes in both directions (66). […] And I do think it is possible that I myself show the patient my trust first and therefore also grant him such a leap of faith in this moment (67). Erm, and in turn the patient acts differently again and then I myself can identify with him more easily and empathize (68). And the more I can identify and empathize with him, the more I can trust him (69). So, I think it's such a circle that constantly reinforces each other, but in which you have to get in first (70).
Since our analysis included few predicates, of which only one combination Ed∧Id was not observed in any possible world, the deduction of only one subtext was possible. The combination Ed∧Id might be expected when David shared the story about a group therapy patient. And, indeed, on request, David reported that at the very beginning of therapy, this patient was very cooperative. It can be assumed that the later incident in the group therapy session destroyed these positive memories, and that is why they were not included in the narrative. However, it was this unspoken background that led to the appearance of a hidden message in the interview.So, I have … I am a bit… It´s a bit of trial and error (71). First, it´s a moral conviction that I have (72). Then, I’ve had positive experiences with it (73). But I also know … or I am afraid… or I know that someday I´ll get in trouble with that because it can certainly happen that I show somebody trusting behavior and the person still might be aggressive and yes, and that then something might happen to me or to others (74).… but that's also, you're right (75). So, I also have patients that I didn´t… erm know or could assess at all … and I trusted them blindly and showed my trust (76). And I mostly, or fortunately always had the experience that they opened up to me or have trusted me and thus I could identify more easily with them and continue to trust them (77).
While David, in the interview, regarded trustful behavior without emotional approximation only as a dubious option, in the course of the subtext analysis, this attitude was identified to be an “ought” statement and there was no alternative. It seems that at the subliminal level, the path of “trial and error” has already been taken. It is noteworthy that the very identification of this attitude and the discrepancies associated with it became possible only after the formal analysis. From a semiotic point of view, the transition from the text of the interview to the subtext marked a change in the theme of violence to the theme of professionalization.
The identified hidden message was most likely intended for internal use. According to Lotman (1990), in autocommunication the bearer of the information remains the same but the message is reformulated and acquires new meaning. This is the result of introducing a supplementary code. The original message is recoded into elements of this code and thereby acquires features of a new message. Perhaps this code was for David the brief information about the attachment theory that he received before interviewing. If so, David's exclamation “…I think it was that closeness!” (Sentence # 33) to the question of what made the situation with the first patient special can be identified as the “eggshell” of this new-born message. The rhetorical device used by David in structuring stories was close to a kind of chiasmus (cross arrangement of the first and fourth, as well as of the second and third signs of key predicates: Ed, ¬Id; ¬Ed, Id). In this regard, there is a link to narrative constructivism, which conceptualizes narrative as a cognitive structure or process (e.g. chi-thinking according to Lissner (2007)) that is to be found inside the mind (Esin et al. 2014; Sparkes and Smith 2013).
However, the message could be addressed to the interviewer as well. In this case, it may rather be seen as co-constructed, or dialogically constructed. The interviewer and the interviewee would then be considered as meaning-makers who use narratives to interpret, direct and communicate life. In turn, the narratives would be assumed to be passed down from people’s social and cultural worlds. With this in mind, narrative constructionism could be a useful analytical approach (Seth et al. 2022; Smith and Monforte 2020).
From contemporary perspective, constructivist and constructionist narrative approaches are not contradictory, but complementary (Riessman 2008). Tellingly, in his study of historical texts, Lotman concluded that they can be coded on at least three levels. First, each text is an utterance in a natural language and, as a result, is organized according to the laws structuring a given language. Second, on a super-phrasal level, each text is subjected to the laws of rhetoric and to the logic of narrative. And third, on the highest level, each text must be ideologically encoded (Tamm 2019).
4 FSA as an extension of QCA
The analysis presented here was first introduced by Klimczak in his PhD thesis as a response to Renner's formalization of Lotman's event outside the context of Configurational Comparative Methods (Rihoux and Ragin 2009). It was formulated first for a small number of predicates without developing a minimization algorithm and controlling for parameters of fit. However, the similarity of FSA to QCA is obvious. In QCA terminology, possible worlds can be roughly interpreted as cases, and predicates with their negations as conditions with membership scores in sets, e.g. “1” and “0”. If we consider Lotman's event model in the frame of the crisp-set QCA, then a fulfilled prescription, e.g. (O(Id)∧Id) in FSA corresponds to the outcome (Id) in QCA. In this framework, perceive set relations between the outcome and the conditions could be interpreted in terms of sufficiency and necessity and represented as a truth table (Table 3). In the expressions ¬Ed ← Id, the condition ¬Ed was necessary (consistency = 1.0, coverage = 0.6) and not sufficient (consistency = 0.60, coverage = 1.0) for the outcome Id. However, the condition Pd in the expression Pd → Id was neither necessary (consistency = 0.33, coverage = 0.33) nor sufficient (consistency = 0.33, coverage = 0.33) for the production of the outcome.
Table 3
Truth table for two conditions (Pd and Ed) and the outcome Id
Row | Pda | Edb | Idc | Consistency (cut off = 1.0) | Cases/Possible worlds |
|---|---|---|---|---|---|
1 | 0 | 0 | 0 | 0.5 | 1, 4, 5, 8 |
2 | 0 | 1 | 0 | 0.0 | 2 |
3 | 1 | 0 | 1 | 1.0 | 6 |
4 | 1 | 1 | 0 | 0.0 | 3, 7 |
The identification of the subtext(s) in FSA formulation is close to finding conditions that lead to the occurrence of an outcome in QCA. Using both classical QMC- and CCubes-minimization algorithms (Duşa 2018), the solution Pd*¬Ed → Id (consistency = 1.0, coverage = 0.33) could be yielded. In this expression, Pd was an insufficient but a necessary part of the condition Pd*¬Ed which was itself unnecessary but sufficient for the outcome Id (INUS condition). This expression was congruent to the causal chain Pd → O(Id) → ¬Ed and the subtext Pd → O(¬Ed) from FSA that, however, did not take into account parameters of fit and therefore considered the condition Pd to be sufficient. Using both QMC and CCubes algorithms, the solution Ed → ¬Id (consistency = 1.0, coverage = 0.33) for the non-occurrence of an outcome was yielded. In all examined formulations, the coverage parameters were low. Using membership move (Pagliarin et al. 2023), it was clear that the low coverage was due to the peculiarity of our calibration of the outcome. A content analysis shows that the processes and results of interventions (Id) in cases 1 and 5, unlike in case 6, were possibly not quite what David would have wanted them to be. This observation could be a basis for a rethinking the calibration.
When adjusted by parameters of fit and minimization algorithms, the technique we are developing can contribute to QCA of qualitative interviews in three ways. First, our extension provides a theoretical foundation for analyzing narrative data based on CCM. Unlike the limited diversity of possible combinations of conditions in sociology or political science, the narrator's ability to create and provide the interviewer with different cases is virtually unlimited. However, text segments that correspond to cases must be combined into a coherent narrative structure, which makes the modal system of reflexive, transitive and symmetric possible worlds (system S5) (Kripke 1959) the most appropriate tool for describing and analyzing text. It can be hypothesized that the subtexts of narrative interviews are structurally different from QCA formulations derived from non-narrative data because subtexts albeit subliminal are inherently created to be understood by the recipient. All this determines the formal uniqueness of truth tables based on narratives and raises the question of an appropriate analysis algorithm, e.g. QMC, eQMC, CCubes (Duşa 2018). Secondly, formalized Lotman's event model with deontic modus ponens or tollens reveals the ethical dimension of the text, which can significantly clear its semantic. Third, adopting Lotman's event model opens up access to his plot theory, which provides a powerful set of tools to acquire various points on the plot scale.
A practical application of our technique as an extension of QCA of qualitative data is to sequentially examine the text for event-inducing and non-event-inducing possible worlds (cases), determination of the relevant predicates (conditions) and primary "ought" statements with fulfilled and unfulfilled prescriptions (outcomes and non-outcomes). Next, a truth table is prepared. Finally, using computerized software, subtext(s) extraction based on the QCA minimization algorithms and deontic modus ponens or tollens is performed. In detail, the analytical process should be guided by recommendations developed by de Block and Vis (2019).
Although the extension discussed here is rather a technique, it is possible to discuss it as an approach and name some relevant rules of application. The methods for conducting interviews are basically the same as described for qualitative research, although the application of some specific techniques, e.g. episodic narrative interviews (Mueller 2019), can be particularly effective. The selection of predicates should be justified in relation to their possible contribution to an event based on sufficient prior theoretical and empirical knowledge. The predicates can be derived from previous studies or from qualitative interviews that already have been conducted based on other approaches (e.g. Grounded Theory). Topics that are suppressed or overemphasized are particularly interesting. Generally, not too many predicates should be selected. A detailed description of their contents should precede the analysis. The predicates should be designed as processes describing different actions of the narrator (process coding). Process coding often, but not always, uses gerunds (“-ing” words) to connote simple observable activity (e.g. dancing, watching TV) as well as more general conceptual (e.g. struggling, surviving) and some mental (e.g. imagining, planning) action (Saldaña 2021). For all described processes, more or less control by the narrator should be possible, so that they could be presented as “ought” statements. Further, the predicates should be selected so that they potentially can be traced for each examined time point of the narration. Although follow-up questions to interviewees are perfectly legitimate, they should be asked very carefully in the form of open questions, as concealment is of diagnostic importance. After initial evaluation of interviews, both, the “ought” statements that could be potentially considered as event-inducing as well as deduced subtexts should be validated in the discussion with interviewed persons (better) or with people from the group representative for the study.
Since the aim of our paper was to provide a preliminary overview of FSA in narrative interviewing, we focused on the crisp-set QCA. For the fuzzy-set QCA, another analytical formulation of FSA with the forbidden operator “F(p)” should be developed. Due to the multivalent nature of qualitative data, studying FSA as an extension to the as yet underutilized multi-value QCA may be promising. In order to conclude on the scientific value of FSA, more analytical data is required.
Declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Ethical approval and consent to participate
Informed consent was obtained from all participants. The ethics permit was issued by Committee of Ethics of Brandenburg Medical School (E-01-20210820).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
